Removal of Emerging Contaminants by Degradation during Filtration: A Review of Experimental Procedures and Modeling

Abstract

Here, we review the efficient removal of organic micropollutants from water by degradation during filtration using specialized bacteria and enzymes. In both approaches, the filter provides essential binding sites where efficient degradation can occur. A model is presented that enables the simulation and prediction of the kinetics of filtration for a given pollutant concentration, flow rate, and filter dimensions and can facilitate the design of experiments and capacity estimates; it predicts the establishment of a steady state, during which the emerging concentrations of the pollutants remain constant. One method to remove cyanotoxins produced by Microcystis cyanobacteria, which pose a threat at concentrations above 1.0 µg L⁻¹, is to use an activated granular carbon filter with a biofilm; this method resulted in the complete removal of the filtered toxins (5 µg L⁻¹) during a long experiment (225 d). This system was analyzed using a model which predicted complete toxin removal when applied at a 10-fold-higher concentration. Enzymes are also used in filtration processes for the degradation of trace organic contaminants, mostly through the use of membrane bioreactors, where the enzyme is continuously introduced or maintained in the bioreactor, or it is immobilized on the membrane.

1. Introduction

The occurrence of organic micropollutants in groundwater and surface waters is of serious concern because of their ubiquity and adverse effects on humans and the environment. These micropollutants comprise a large list of organic chemicals, such as flame retardants (brominated diphenylethers), alkylphenol ethoxylate surfactants (nonylphenol), chloroalkanes and phthalates, certain herbicides, pharmaceuticals (PhACs), personal care products, etc. Some of these micropollutants have been regulated as priority hazardous substances, but others are denoted as emerging contaminants (ECs) due partly to the inherent difficulties in analyzing these chemicals.

Their presence in the environment is derived from diverse sources, such as the improper use of chemicals associated with agricultural practices, leaching from urban and industrial solid waste dumps or septic tanks, untreated domestic discharge, and in particular, inefficient treatment of urban effluents, which is the main contributor to their presence in water bodies. Key examples of how this pollution is present across all continents have recently been reviewed, indicating the need for upgrades to wastewater treatment plants [1]. This upgrade includes introducing new utilities to the existing ones, including some recently developed treatment processes, such as membrane bioreactors, advanced oxidation processes, filtration using new sorbent materials, etc. [2,3,4].

The focus of this mini review is to present efficient and economical technologies for the removal of organic micropollutants from water by degradation during filtration using (1) specialized bacteria or (2) enzymes. In both approaches, the filter provides essential binding sites where efficient degradation can occur. The examples of removal of several micropollutants from water will cover three ranges of concentrations:

(i) 1–500 ng/L. In this category, the removal of molecules which affect taste and odor (T&O) is important in the treatment of both drinking water and water in fish ponds. We will elaborate on the case of fish ponds.

(ii) 1–200 µg/L. Drinking water treatment plants which utilize lake water are increasingly encountering the need to remove cyanobacteria toxins from water during harmful algal blooms. The cyanotoxins are life-threatening both for humans and animals, and their concentrations in drinking water should not exceed 1 µg/L.

(iii) 0.001–10 mg/L. In this range, we will mainly discuss the removal of pharmaceuticals. The limits on their concentrations in drinking water vary from 0.1 µg/L to several mg/L.

Section 2 presents differential equations which describe the process of degradation during filtration. The tables provide numerical values which demonstrate the effect of flow velocity; rate constants of degradation, adsorption, and desorption, pollutant concentration; and filter length for filtration times ranging from 1 h to several days. The equations exhibit the establishment of a steady state during which the emerging concentrations of pollutants and their sorbed amounts in the filter remain constant. Section 3 presents methods for the removal of taste and odor molecules from water in fish ponds and presents experimental and calculated values of their removal in the presence of degrading bacteria or under inhibition of bacterial activity by azide. The potential of model calculations to yield predictions for the concentrations of taste and odor molecules during 6 weeks of filtration is demonstrated. Section 4 addresses the issue of harmful cyanobacteria blooms and reviews an efficient procedure for the removal of cyanotoxins by specialized bacteria, which form biofilms in a GAC filter. The experimental results demonstrate an efficient removal of cyanotoxins from water over 225 d using this procedure. An adequate simulation of the experimental results is presented. A tentative projection based on modeling is presented for extending the treatment to higher toxin concentrations. In Section 5, the degradation of pharmaceuticals using enzymes is described, which preferentially uses membrane filtration processes and, occasionally, fluidized and bed filtration systems.

2. Model Equations and Analysis: Removal of Emerging Contaminants by Degradation during Filtration

The model presented in this section can simulate and predict the kinetics of filtration for a given pollutant concentration, flow rate, and filter dimensions and can facilitate the design of experiments on laboratory and pilot scales. The filtration (without degradation) model [5] has yielded simulations and predictions for the removal of more than 30 contaminants [6,7,8,9,10,11] and yielded predictions for a pilot-scale study [12]. An extended model for degradation during filtration was described in [13].

The modeling of filtration data employs equations which consider convection and adsorption/desorption [5] in a system that includes several solutes. Equation (1) is for a solution which includes only one pollutant, whose molar concentration is C (X, t), where X is a coordinate along the filter length and t is the time.

The top and bottom of the filter are at the coordinates X = 0 and X = L, respectively. The pollutant concentration at the inlet, C₀ is constant, i.e., C (X, t) = C₀ for X ≤ 0, where t denotes time. There is also an unpublished version of the model which enables the measurement of changes in C₀ and flow rates during the course of filtration.

The derivative of C (X, t), with respect to time, is given by:

dC (X, t)/dt = −v·∂C (X, t)/∂X − C₁ C (X, t)·R (X, t) + D₁·RL (X, t),

(1)

in which R (X, t) and RL (X, t) are the concentrations of free and occupied sites, respectively.

In Equation (1), v (cm/min) denotes the flow velocity in the filter pores,

v = Q_v/(A·f),

(2)

In which Q_v is the flow rate (volume/time), and f is the fraction of pore volume in the filter.

R (X, t) = R₀ − RL (X, t)

(3)

C₁ is the forward rate constant of adsorption (M⁻¹min⁻¹), and D₁ (min⁻¹) is the rate constant of dissociation. R₀ is the total molar concentration of adsorption sites in the filter.

Degradation or inactivation of the pollutant is assumed to occur due to the bacteria in the biofilm in the filter or due to the enzymes in the filter according to:

d RL (X, t)/dt = C₁·C (X, t)·R (X, t) − (D₁ + kd)·RL (X, t),

(4)

in which kd (min⁻¹) is the rate constant of solute (pollutant) degradation or inactivation by the enzyme or bacteria. For (computationally) testing the mass conservation of the solute, the concentration of degraded solute is given by:

DB (X, t), obtained from

d DB (X, t)/dt = kd·RL (X, t)

(5)

The numerical solution of the above equations for filtration was described in Nir et al. [5,12] and was extended for degradation in the case of a single solute in [13].

In all cases studied, the solutions yielded a steady state, i.e., the emerging concentrations of the pollutant and its adsorbed concentrations remained constant, whereas the pollutant was degraded at a constant rate. The calculations require knowing the quantities C₀ (given) and R₀, and the rate constants C₁, D₁, and kd. The first three parameters can be obtained from the analysis of the filtration results in which the filter includes an azide solution, which inactivates the degrading bacteria, or using an inactivated enzyme. Then, the parameter kd, the rate constant of degradation of each pollutant by the bacteria or enzyme (e.g., laccase), can be determined from the filtration of each pollutant through a filter, which includes the bacteria or immobilized active enzyme.

Table 1. Filtration of pollutant (C₀ = 5 × 10⁻⁶ M) with or without degradation ^a through column with length of 20 cm and radius of 5 cm, at a flow rate of 20 mL min⁻¹.

		kd = 0	kd = 0.01 min−1		kd = 0.001 min−1
Time (h)	Volume (L)	C/C0	C/C0	% degrad.	C/C0	% degrad.
1	1.2	0.025	0.025	18	0.025	2.1
5	6	0.075	0.039	63	0.068	12.2
10	12	0.16	0.041	78.7	0.123	22.5
24	28.8	0.45	0.041	88.7	0.232	41
48	57.6	0.83	0.041	92.3	0.284	54.6
96	115	0.993	0.041	93	0.289	62.8
120	144	1	0.041	94.5	0.289	64.5

Notes: ^a the fraction of pore volume out of the total volume was 0.5. The model parameters were R₀ = 0.002 M; C₁ = 50 M⁻¹min⁻¹; D₁ = 0.002 min⁻¹.

gives an example of the filtration of a pollutant through a filter under three given situations: kd = 0 (without degradation), 0.01, and 0.001 min⁻¹. A filter with a volume of 1570 mL includes 400 g of adsorbing material. The results of filtration are expressed by C/C₀, i.e., the ratio between the emerging concentration C and the initial concentration C₀. If the allowed value of the given pollutant for being spilled into the environment corresponds to 0.45 C₀, then the capacity of this filter is 28.8 L, which amounts to 28.8/400 L/g. When regenerated by an optimized procedure based on microwave heating, as in [14], the capacity can be tripled to 0.22 (m³/kg), which is rather low. When a procedure augmented by degradation with kd = 0.01 min⁻¹ is employed, the calculated results in Table 1 indicate that after 10 h of operation, the value of C/C₀ is just 0.041 or C = 0.041C₀ up to 5 d, and in fact, for longer times, implying a very large capacity. For the lower value, kd = 0.001 min⁻¹, Table 1 indicates that at times of 4 d and beyond, the value of C/C₀ remains 0.289 (Table 1), which also corresponds in this case to a very large capacity.

Table 1 indicates that degradation during filtration implies an establishment of a steady state during which the value of the concentration of the emerging pollutant remains constant. Inspection of other quantities calculated by the model indicates that the concentrations of the adsorbed sites remain constant after initiation of a steady state. Thus, during a steady state, the pollutant is being degraded, and its cumulative percent of degradation keeps increasing. The combination of parameters dictates the emerging concentrations of the pollutant, which can reach practically zero or exceed the allowed values. Column 6 in Table 1 shows that terminating the filtration after 24 h would not permit the system to reach a steady state. Hence, increasing the time of the experiment is recommended. This requirement may be moderated by increasing the length of the filter without a corresponding increase in the flow rate. An example is provided in Table 6.

Table 2. Filtration of pollutant (C₀ = 5 × 10⁻⁶ M) ^a through a column with a length of 40 cm and radius of 1 cm, at a flow rate of 20 mL∙min⁻¹.

Time (h)	Volume (L)	Input Mole	Sorbed (Mole)	Sorbed (%)	Degraded (%)	C/C0
1	1.2	6 × 10−6	3.46 × 10−6	57.7	18.7	0.228
5	6	3 × 10−5	7.17 × 10−6	23.9	52	0.249
10	12	6 × 10−5	7.48 × 10−6	12.4	63	0.251
24	28.8	1.44 × 10−4	7.49 × 10−6	5.2	69.9	0.251
48	57.6	2.88 × 10−4	7.49 × 10−6	2.6	72.4	0.251
96	115.2	5.76 × 10−4	7.49 × 10−6	1.3	73.7	0.251

Notes: ^a the fraction of pore volume out of the total volume was 0.5; R = 1 cm. The model parameters used were R₀ = 0.005 M; C₁ = 100 M⁻¹min⁻¹; D₁ = 0.001 min⁻¹, kd = 0.01 min⁻¹.

presents another situation with the same pollutant concentration and flow rate as in Table 1, but the flow velocity was 25-fold higher since the radius was 5-fold smaller. Table 2 presents the increase in the cumulative number of sorbed moles of pollutant with time, until the steady state is reached, similar to the values of emerging pollutant concentrations or C/C₀. On the other hand, the cumulative percentage of pollutant degradation continued to increase with time at the expense of its sorbed concentrations. Thus, after 96 h, the (cumulative) percentage of sorbed pollutant was just 1.3. This demonstrates the relative importance of the degradation process in the removal of a degradable pollutant.

3. Removal of Taste and Odor Molecules from Water in Fish Ponds

Taste and odor (T&O) removal is occasionally needed in the treatment of drinking water. Here, we will elaborate on the solution of this problem in fish ponds. Currently, aquaculture accounts for half of the world’s fish supply [15]. Recirculating aquaculture systems (RASs) [16] enable an appreciable conservation of water resources. With the development of high density-operated culture systems, the problem of T&O or “off-flavored” fish has increased [17].

Geosmin and 2-methylisoborneol (MIB), which are considered to be the main molecules associated with the off-flavor in fish ponds, are produced as secondary metabolites by various microorganisms. They produce an earthy–musty taste and odor. Due to their hydrophobic nature (log K_ow is 3.57 and 3.31 for geosmin and MIB, respectively), they accumulate in the fatty tissues of fish and render the fish unmarketable [18]. These compounds have been observed in conventional fish farms [19], as well as in fresh water [20] and marine [21,22] RAS facilities. People reported an earthy or musty smell and flavor when MIB or geosmin concentrations exceeded 10 ng/L [23]. Thus, for water containing MIB and/or geosmin, effective T&O control can only be achieved by employing a technology that can reduce the MIB and geosmin concentrations to very low levels. Pre-marketing depuration (purging) of tainted fish, aimed at reducing the effect of geosmin and MIB, is currently a common treatment; however, it consumes large amounts of clean water and results in weight loss of the depurated fish [24,25].

Studies conducted in a zero-discharge RASs indicated that the removal of waterborne geosmin and MIB can be achieved using a combination of adsorption and biodegradation by sludge in the digestion basin of the system [20]. In a subsequent study [26], it was found that an initially rapid MIB adsorption onto sludge particles was followed by a recharge of the sludge’s adsorption sites through degradation of MIB by an endemic bacterial consortium. Under steady-state conditions, biodegradation was found to account for >99% of the overall MIB removal [26]. A numerical model was developed to describe the combined adsorption/biodegradation of MIB in the sludge [26]. In a later study, the enrichment of the sludge derived from the same RAS above with geosmin and MIB resulted in a significantly increased abundance of 12 distinct bacterial species [27].

Here, we largely describe selected results from [13] on the removal of MIB, which is the least hydrophobic of the two off-flavor molecules, by employing the model described in Section 2 of this review, and a few calculated results beyond those in [13] are provided. We present the results for the optimized hydrophobic carriers which were included in the filter. The calculations of MIB removal from water using a combination of adsorption and biodegradation during filtration employed the same parameters as in [13].

The filter included hydrophobic spherical carriers of about 4 mm in diameter. The carriers, which were designed to provide sufficient mechanical strength, were developed by modifying alginate (with a molecular mass of 60–70 kDa). The freshly prepared carriers contained 30% soy oil, which made them hydrophobic.

The filters (denoted as reactors) employed were operated in an upflow mode (volume of 1.3 L; diameter of 7.5 cm). The carriers (50 g,

Table 3. Removal of MIB from recirculating aquaculture water (TOC content: 20–50 mg L⁻¹) by filtration with carriers packed in laboratory-scale upflow reactors in the presence (+azide) and absence of azide (−azide). Experimental data represent mean values of duplicate runs. The flow rate was 5 mL/min. The carrier load was 50 g ^a.

	% MIB Emerging from Filter
	Experimental		Calculated		MIB Degraded
Time	+Azide	−Azide	+Azide	−Azide	Calc. (%)
1 h	33.1 ± 1.7	30.6	18.7	18.2	27.7
2 h	NA b	NA	22.3	18.9	48.0
10 h	NA	NA	48.2	19.0	73.0
24 h	77.8 ± 1.2	19.5 ± 0.9	77.8	19.0	78.0
48 h	96.5 ± 4.0	20.3 ± 1.0	96.8	19.0	79.4
96 h	100 ± 0.1	18.9 ± 0.9	99.9	19.0	80.1
168 h	100 ± 0.1	19.5 ± 0.7	100	19.0	80.4
2 w	-	17.4 ± 0.6	-	19.0	80.7
3 w	-	18.0 ± 1.1	-	19.0	80.8
4 w	-	15.1 ± 1.9	-	19.0	80.8
5 w	-	17.8 ± 0.5	-	19.0
6 w	-	19.7 ± 0.3	-	19.0

Notes: ^a the length L of filter (length of active layer of carriers) was 3.75 cm; C₀ = 1000 ng L⁻¹ (5.95 × 10⁻⁹ M); R₀ = 4.4 × 10⁻⁶ M; the fraction of pore volume (f, Equation (2)) was 0.78; C₁ = 12,000 M⁻¹∙min⁻¹; D₁ = 0.002 min⁻¹; kd = 0.025 min⁻¹ (degradation rate constant is zero in the case of (+azide)). ^b not available.

) were packed in small net bags and placed on top of a “false bottom” about 5 cm above the bottom of each reactor. The reactors were fed with crude RAS water containing 20–50 mg/L of TOC, spiked with 1000 ng/L of MIB. The filtration used a peristaltic pump with Tygon tubing. The flow rates were set at 0.3 L/h (5 mL/min). In order to be able to distinguish between adsorption and bacterial degradation, the experiment also included a filter amended with 400 mg/L of sodium azide.

The results of the experiments in which azide was included (Table 3, column 2) correspond to filtration without degradation. In this case, no removal of MIB was observed after 4 d, whereas after 2 d, the concentration of the solution emerging from the filter reached 96.5% of the initial MIB concentration. These results enabled us to determine the parameters R₀, C₁, and D₁ (see Section 2). The calculated values of MIB concentration in the emerging solution in column 4 of Table 3 adhered well to the experimental values in column 2 from 1 d to 7 d. The value of the rate constant of degradation, kd, was determined by simulating the results in column 3 (−azide) up to 4 d, and the fit of the calculated values (column 5) was very good. The experimental values clearly indicate the establishment of a steady state at day 2, during which the concentration of MIB in the emerging solution was about 20% of the initial value. The model calculations suggest an initiation of a steady state after 10 h. The calculated values yielded predictions for the experimental values obtained between 4 d and 42 d. Inspection of column 3 indicated that the experimental values of the initial MIB concentrations in the emerging water from the filter fluctuated around the predicted value of 19% of the initial concentration. Considering the initial concentration of 1000 ng/L, this amounts to a rather large value, which was used for analytical convenience. The calculations in this article showed that using a smaller initial MIB concentration value of 100 or 50 ng/L will yield the same results as in Table 3 in terms of percent MIB in the emerging solution relative to the initial concentration. Hence, for passing through the filter a crude RAS water containing 20–50 mg/L of organic matter spiked with 100 ng/L of MIB, the emerging water would include 19 ng/L of MIB.

As was also shown in Section 2, elongation of the filter, while keeping all other conditions identical, can lower the emerging MIB concentration. We calculated the results for two cases in which the filter length was (i) 10 cm or (ii) 20 cm (rather than 5 cm in Table 3). At a steady state, the values of C/C₀ (expressed as % of MIB emerging from the filter) were (i) 4.8% or (ii) 0.18%, rather than 19% in Table 3, and the corresponding emerging MIB concentrations when using an initial MIB concentration of 100 ng/L were (i) 4.8 or (ii) 0.18 ng/L. This implies that a large capacity can be expected for a filter (reactor), which can maintain a low MIB concentration in the water of a fish pond operated by an RAS.

4. Removal of Cyanotoxins from Water

4.1. Removal of Cyanobacteria and Cyanotoxins by Filtration

Cyanobacteria, such as Microcystis aeruginosa (Chroococcales) and Aphanizomenon ovalisporum (Nostocales), produce toxins that are harmful to humans and animals. According to [28], there are upper limits for these toxins in drinking water (e.g., MC-LR < 1 μg/L). Cyanobacteria blooms are an increasing worldwide health hazard and are denoted as harmful algal blooms (HABs), which can cut off the main supply of drinking water to millions of people, despite the existence of advanced water treatment plants. The leading view is that eutrophication is a major driving force of the intensification of cyanoHABs worldwide, in addition to global warming [29,30,31,32].

The efficient removal of Aphanizomenon and Microcystis cells by filtration was demonstrated and modeled in [7] by employing filter columns including a granulated micelle–clay composite [12]. The granulated complex is positively charged to about half of the CEC (cation exchange capacity) of the clay (or clay mineral) and includes hydrophobic domains [33]. In [34], it was shown that the removal of Microcystis cells by filtration might be an economical solution for cell concentrations below 10⁵ per mL if a pre-treatment with flocculation as described in [35,36,37] is used. However, this procedure cannot be applied in the case of HABs. In this section, we will discuss how to augment a water treatment plant, whose last treatment stage would deal with the removal of cyanotoxins to the level required by the regulations. In the presence of very large concentrations of cyanobacteria cells, it is assumed that killing these cells can be accomplished by the addition of small concentrations of a salt of a quaternary cation, such as octadecyltrimethylammonium (ODTMA). The required concentration of the quaternary cation ODTMA for killing large filamentous aphanizomenon cells (2700/mL) for a 20 min exposure was 0.25 µM (75 µg/L), whereas for the same cellular volume of microcystis cells (5.5 × 10⁶/mL) for a 40 min exposure, 6 mg/L of ODTMA was required. In fact, the above values may be lower for longer exposure times. However, the needed ODTMA concentrations for killing other bacteria, such as E. coli, or for reducing Total Bacteria Counts are significantly larger. Another procedure that is employed is based on the addition of hydrogen peroxide (H₂O₂) [38,39].

Adsorption of several cyanotoxins by powdered activated carbon [40] and by GAC and advanced oxidation processes (AOPs) were reported in studies on their removal from water [28,41]. Sukenik et al. [7] demonstrated adsorption of the toxins microcystins (MCs) and cylindrospermopsin (CYN) from a suspension by 1 g/L of a micelle (ODTMA)–clay complex. This was followed by [42], who determined experimentally and modeled the removal of several microcystins by filters, including a granulated micelle (ODTMA)–clay (bentonite) complex. The MC congeners MC-LR, MC-YR, MC-WR, and MC-3aspWR were removed more efficiently than the more positively charged ones (MC-RR and MC-3aspRR). The model (see Section 2) was first applied to simulate and predict the removal of MC-LR, which was the most potent toxin. Subsequently, the parameters describing the other toxins were determined from the results from the filtration of MC-LR with another toxin. One conclusion of this study was that a filter with the granulated complex can be applied in the removal of microcystins from drinking water. The calculations in [42] showed a capacity of 4.3 m³/kg of the complex for the removal of MC-LR from a 5.5 µg/L solution (for an emerging toxin concentration below 1 µg/L), whereas for a 10 µg/L solution, the capacity corresponded to 3 m³/kg of the complex. These capacities are not economical. In Section 4.2, we will describe a scheme in which the incorporation of the micelle (ODTMA)–bentonite complex can be useful.

The results in [42] indicated a more efficient removal of the main toxins MC-LR and MC-YR by filtration using a granulated ODTMA–clay complex compared to GAC, whereas the more positively charged congener MC-RR was removed more efficiently by GAC.

4.2. Removal of Cyanotoxins by Degradation during Filtration

Whereas we focused on the removal of microcystins, it is instructive to mention a study [43] which concluded that the biodegradation of the toxin CYN was only evident in water supplies that had a history of toxic Cylindrospermopsis blooms. We will first describe the experimental design and results of Wang et al. [44]; next, our calculated results (simulations and predictions) will be presented, followed by a suggestion for an extension of the experimental design.

A.

Laboratory experiments on removal of microcystins [40]

The authors of [44] studied the removal of the toxins MC-LR and MC-LA. The water was provided from a Myponga Water Treatment Plant (WTP) in South Australia. The water used had previously undergone all the regular treatment stages in the plant but was removed prior to the final chlorination step. The water included, on average, 6 mg/L of dissolved organic carbon (DOC), whose absorbance at 254 nm was 0.1 cm⁻¹. The water was spiked with a 5 µg/L toxin solution. The coal-based GAC used contained a high proportion of micropores and mesopores, which was considered to be advantageous for the adsorption of microcystins. Three laboratory column filters were operated in parallel (in duplicate): (i) a GAC column, which recorded the removal of the toxin due to adsorption and degradation during filtration. In this case, the involvement of bacteria in the degradation of cyanobacteria toxins was studied in detail. At the completion of the filtration experiment, 5 g (wet weight) of material was removed from the GAC column and after vortexing, the bacteria within the biofilms were detached and collected for bacterial inoculation of bioreactors at 7.6 × 10⁷/mL, which were incubated under aerobic conditions. The study also investigated biodegradation in batch experiments, which demonstrated the effect of temperature on toxin degradation. At 40 °C, no degradation was observed, whereas optimal rates of degradation occurred at 25 °C and 30 °C. (ii) A column filled with sterile GAC was used to record adsorption during filtration. (iii) A filter filled with sand, where each column was 15 cm long and 2.5 cm in diameter, was operated at a flow rate of 4.9 cm³/min.

The results of filtration by the GAC filter indicated that the toxins were completely removed during the first few days, then some breakthrough was observed (not exceeding the allowed level for MC-LR) up to day 38, after which complete removal was observed until the end of the experiment after 225 days.

The filtration through the sand filter (see also Ho at al. [45]) did not yield any toxin removal during the first 75 days, but it was complete between 220 and 225 days, due mostly to degradation by specialized bacteria which could settle on the sand granules. The filter with sterile GAC resulted in the complete removal of the toxin MC-LR during the first 5 days, 80% removal after 80 d, and about 70% removal on average at later time points. Clearly, the results are very impressive and show a capacity of at least 50 m³/kg GAC for a 5 µg/L solution of MC-LR or MC-LA, since the operation could continue beyond 225 days.

The last sentence in the Abstract of [44] states that “Up to 70% removal of MC-LR was still observed after 6 months of operation of the sterile GAC column, indicating that absorption still played a vital role in the removal of this toxin”. In contrast, in Section 2 and Section 3 of the current article, we emphasized that most of the removal of pollutants is in the long run due to degradation during filtration. It should be noted that the sterilization of the GAC was conducted once a week in order to avoid the development of toxin-degrading bacteria in the filter. The process of sterilization of the GAC involved autoclaving at 121 °C. In [14,46], the heating of the used filter material was carried out as an efficient regeneration, which restored most of its adsorption capacity. During the first stages of removal via degradation during filtration, most of the removal occurs by adsorption rather than by degradation. Hence, the sterilized GAC (after the first few weeks) may have an advantage until the contribution from degradation dominates. We noted in Section 2 that, when the steady state is established, there is no direct measurable contribution to pollutant removal from adsorption, albeit we expect that the bacteria which degrade the toxins primarily act on adsorbed molecules.

B.

Simulation of results in (A) and a proposed treatment for larger toxin concentrations

Here, we present calculations of a simulation of the results in [44]. In the absence of data on toxin removal in the presence of azide, as in Section 3, we estimated the parameters R₀, C₁, and D₁ by considering results in [42] and in Kummel et al. [47], and tested whether by setting kd = 0, the calculated removal during the first 5 days is complete. The outcome of calculations without degradation is shown in

Table 4. Calculated filtration outcome for a 5 µgL⁻¹ MC-LR solution by a GAC column as in [44], i.e., 15 cm long and a diameter of 2.5 cm at a flow rate of 4.9 mL/min in the absence of degradation ^a.

Time (d)	C/C0	Emerging Concentration, µg L−1
2	0	0
3	0.045	0.03
5	0.109	0.5
6	0.209	1.04
8	0.476	2.4
10	0.729	3.6
16	0.988	4.9

Note: ^a the parameters employed in the calculations were R₀ = 0.005 M, C₁ = 500 M⁻¹min⁻¹, D₁ = 0.0015 min⁻¹.

. The table indicates that, during the first 5 days, the filtration calculations which use the chosen parameters indicated the complete removal of the toxin, whereas beyond 16 days, adsorption without degradation did not show any removal.

In

Table 5. Calculated filtration with degradation for a 5 µg L⁻¹ MC-LR solution by a GAC column (same as that in Table 4, but kd = 0.00035 min⁻¹) ^a.

Time (d)	C/C0	Emerging Concentration (µg L−1)	Percent Degradation (Cumulative)
2	0.00084	0.0047	37
4	0.008	0.04	56.9
5	0.014	0.07	63.4
6	0.02	0.1	68.3
10	0.03	0.15	79.4
20	0.031	0.155	88.1
34	0.031	0.155	91.4

Note: ^a the parameters employed in the calculations are the same as those in Table 4, but a degradation rate constant, kd = 0.00035 min⁻¹, was added.

, the calculated results employed the same values of the parameters as in Table 4, but degradation was considered by setting kd = 0.00035 min⁻¹. The results indicate that, for the first 5 days, the concentration of the emerging toxin was practically zero. The establishment of a steady state occurred on day 20; the value of the emerging toxin concentration was 0.155 µg/L, the same as on day 34, and was maintained until the end of the experiment on day 225 and would have been potentially maintained longer if the experiment had continued.

There is still a question of how well the treatment may work if it is extended to the removal of cyanotoxins at a higher initial concentration. To answer this, we considered a solution of MC-LR at a concentration of 50 µg L⁻¹ and used the same filter as in Table 5. The calculations in this article showed that the values of C/C₀ were similar to those in Table 5, albeit slightly larger. During the first two days, the use of the filter yielded emerging concentrations in line with the regulations, e.g., on day 2, C/C₀ = 0.0047, which, after multiplication by C₀, yielded a value of 0.23 µg L⁻¹, but on day 4, the value of C exceeded the allowed values.

Table 6. Calculated filtration with degradation for a 50 µg L⁻¹ MC-LR solution using a GAC column (same as that in Table 5 but the filter length was 30 cm) ^a.

Time (d)	C/C0	Emerging Concentration (µg L−1)	Percent Degradation (Cumulative)
1	0	0	21.4
14	8.3 × 10−4	0.041	85.8
18	9.6 × 10−4	0.048	88.9
24	9.8 × 10−4	0.049	91.7
30	9.8 × 10−4	0.049	93.3
34	9.8 × 10−4	0.049	94.1

Note: ^a the parameters employed in the calculations were the same as those in Table 5.

shows that elongating the filter to 30 cm can solve this problem. In this case, the emerging toxin concentration on all days after day 24 was 0.049 µg L⁻¹, which implies that the filter capacity is very large, as long as the environment of the degrading bacteria does not worsen.

C.

Addition of a micelle–clay filter

In the previous sections, an economical procedure was described for the removal of cyanotoxins from drinking water, using a filter filled with GAC and with a biofilm of toxin-degrading bacteria. The same element can be added to a water treatment plant. The results in [44] demonstrated an inactivation of the bacteria which degrade microcystins at a temperature of 40 °C. It can be assumed that degradation of toxins by bacteria can also be endangered due to an unexpected spill of toxic material which the water treatment plants are not designed to remove very efficiently. We propose the design of an innovative tandem filtration system consisting of a GAC filter followed by a granulated micelle (ODTMA)–clay (bentonite) complex. It has been previously found that the capacity of a micelle–clay filter is rather insensitive to temperature [34].

As was mentioned in Section 4.1, the micelle–clay filter can remove cyanotoxins for a limited period of 2 days, which means that their removal can proceed for a period of time until the GAC filter is replaced. The second filter can also be useful in removal of bacteria or viruses which cannot be effectively removed by the GAC; this also includes the bacteria that detached from a biofilm in a GAC filter. In addition, this filter was shown to be effective in the removal of quite a few ECs.

5. Removal of ECs Using Enzymes

The use of enzymes for micropollutant degradation is based on their inherent properties such as high selectivity, low cost compared to other chemically catalyzed processes, lower susceptibility to inhibition by toxic substances, and effectiveness in a large range of concentrations [48]. Ligninolytic enzymes from white-rot fungi, particularly extracellular oxidoreductase enzymes, have been used because of their relative non-specificity toward organic chemicals and high potential to oxidize them [49]. The most widely used enzymes are laccases and peroxidases. Laccase is a multi-copper enzyme in which the oxidation of the substrate is accompanied by the reduction of molecular oxygen to water. Peroxidases are heme enzymes in which the catalytic cycle involves the use of H₂O₂ or organic hydroxyperoxides as a co-substrate. They comprise a broad group, with the most common being manganese peroxidase (MnP), lignin peroxidase (LiP), and versatile peroxidase (VP). Other widely studied peroxidases from other sources are horseradish peroxidase (HRP) and soybean peroxidase (SPB) [50].

These enzymes catalyze the oxidation of a wide variety of organic compounds, including phenols, methoxy-substituted phenols, aminophenols, diamines, aromatic amines, and some inorganic ions [51]. In certain cases, especially in the oxidation of non-phenolic substrates, the range of molecules which can be oxidized can be extended by the use of mediators. Mediators are small molecules of natural or synthetic origin, such as violuric acid, syringaldehyde (SA), 1-hydroxybenzotriazole (HBT), 2,2’-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), etc., that are oxidized by the enzyme, generating highly reactive radicals that attack and oxidize the target molecule [52,53] (Figure 1). As an example, the degradation of the recalcitrant pharmaceutical carbamazepine by laccase reached values up to 60% in the presence of HBT, whereas its degradation in the absence of the mediator was negligible [54]. Almost complete degradation of tetracycline and sulfonamide antibiotics was obtained using SA as a mediator vs. rates of about 60% with laccase alone; however, no enhanced removal of quinolone antibiotics was observed [55].

The selection of an enzyme system for the degradation of emerging contaminants (ECs) will depend on the redox potential of the enzyme, the mediator used, and the experimental conditions of the reaction medium. LiP is a powerful oxidizing agent with a redox potential of +1.2 V at pH 3 and does not require mediators in the oxidation of non-phenolic substrates (compare this to redox values of +0.8 V for MnP and fungal laccases); however, the LiP redox potential decreases sharply with pH in contrast to laccase, whose optimum activity is between pH levels of 3.0 and 6.0 [56].

Under experimental conditions in which laccase was not able to oxidize sulfonamide and tetracycline antibiotics, the use of mediators such as SA and HBT almost completely eliminated the presence of the parent compounds in solution, with the laccase–SA system showing a better performance compared to using HBT as a mediator [55]. The use of ABTS instead of SA provided a fast transformation of sulfonamide antibiotics [57]. These features are related to the underlying oxidation mechanisms: (i) the ABTS mediator undergoes electronic oxidation by the enzyme to the cation radical ABTS, which diffuses from the catalytic pocket, oxidizes the substrate, and transforms back to ABTS (electron transfer (ET) pathway); (ii) HBT and SA operate through a hydrogen atom transfer (HAT) pathway, with their monoelectronic oxidation generating, respectively, N-oxyl and phenoxy radicals, which target the substrate by extracting a hydrogen atom, and reverting back to HBT and SA. The presence of methoxyl groups in SA increases the electronic density of the phenoxyl moiety, reducing the redox potential of the phenoxy moiety and facilitating its oxidation by the enzyme. The differences in the redox potential of the mediator once it is oxidized and the substrate are not relevant due to the HAT route, unlike the case with ABTS, which forms a more stable radical in solution [58]. The redox potential in the ABTS–laccase system was slightly higher than that with SA–laccase, yielding reduced half-lives in the degradation of sulfonamides [57].

Several studies have focused on the use of crude enzyme extracts and whole fungal cells to reduce the operational costs by reducing the purification step to obtain oxidoreductase enzymes. Tran et al. [59] studied the removal of 10 PhAcs using whole fungal cultures of T. versicolor and after filtration (“crude enzyme extracts”), showing complete removal of three of the PhAcs (diclofenac, naproxen, and indomethacin); however, the crude enzyme extract performed worse with the other PhAcs (5–25% removal) compared to whole fungal cells (>70% removal). Li et al. [60] obtained a 90% removal of naproxen with the crude enzyme extract of P. chrysosporium, which was not achieved until a 3.5-fold-longer reaction time had elapsed compared with that of the whole cell culture. In contrast, carbamazepine was not eliminated using the crude enzyme extract, while the use of the whole fungal culture yielded a 30% removal. Elimination values of 94% of carbamazepine were reported using whole cultures of T. versicolor, revealing the role of intracellular enzymes, such as the cytochrome P450 enzyme system, in the degradation of xenobiotics [61]. Nguyen et al. [62] indicated that the removal of ECs is compound-specific, and differences in degradation by the crude enzyme extracts and the whole cell culture can be rationalized on the basis of biosorption vs. biodegradation, the role of intracellular enzyme systems, and mycelium-associated enzymes. In certain cases, these authors obtained better removal using a mediator-amended extracellular extract compared to the whole cell culture.

Irrespective of the use of mediators, the oxidation of the substrates by enzymes will be strongly dependent on their chemical structure. The presence of electron-withdrawing groups (EWG), such as amide, carboxylic, halogen, and nitro groups, generated a lower electron density in the aromatic ring, decreasing the reactivity of the enzyme to oxidize the substrate molecules. On the contrary, the presence of electron-donating groups (EDGs), such as amine, hydroxyl, alkoxy, alkyl, and acyl groups, enhanced the reactivity and promoted oxidation [63]. The PhAcs carbamazepine, diclofenac, and ibuprofen were resistant to enzymatic degradation due to the presence of EWGs in their structure [64]. On the contrary, antibiotics with EDGs were very susceptible to degradation (e.g., tetracycline, ofloxacin, sulfamethoxazole, metoprolol, gemfibrozil, trimethoprim, acetaminophen, indomethacin, etc.) [65,66,67].

An assessment of the effectiveness of the enzymatic degradation of ECs is difficult due to the diversity of experimental conditions used along with the type of enzyme and its source (

Table 7. Degradation of ECs by oxidoreductase enzymes.

Enzyme (Producer)	EC(s) 1	EC Concentration	Enzyme Concentration	Degradation (%)	Reference
Crude MnP (Pleurotus ostreatus)	CBZ	10 mg/L	7.7 U/L	99.7	[68]
Crude LiP (Phanerochaete chrysosporium)	CBZ, DC	5 mg/L 5 mg/L	180 U/L	100 (pH 3.0–4.5)	[69]
Laccase (Trametes versicolor)	CBZ	1 mg/L	60 U/L	95	[70]
HRP laccase (Trametes versicolor)	Steroid estrogen mixture (EE, E2, E3)	100 ng/L of each	8 × 103–10 × 103 U/L 20 × 103 U/L	100 100	[71]
HRP	DC	25 mg/L	6.4 × 103 U/L	47	[72]
Crude MnP (Trametes maxima)	DC SMX IND GFB BZF	2 mg/L of each	387.6 ± 67.4 U/mg	90.2 72.6 60.8 43.4 32.6	[73]
Crude MnP (Phanerochaete chrysosporium)	TC OTC	50 mg/L	40 U/L	72.5 84.3	[74]
Laccase (Pycnoporus sanguineus CS43)	NP TCS	10 mg/L	100 U/L	94 92	[75]
Laccase (Pleurotus eryngii)	LV	5 mg/L	2 × 103 U/L	97	[76]

Notes: ^1. Abbreviations: CBZ: carbamazepine, DC: diclofenac, E2: 17ß-estradiol, E3: estriol, EE: 17α-ethinyliestradiol, SMX: sulfamethoxazole, IND: indomethacin, GFB: gemfibrozil, BZF: bezafibrate, TC: tetracycline, OTC: oxytetracycline, NP: nonylphenol, TCS: triclosan, LV: levofloxacin.

).

5.1. Use of Enzymes in Filtration Processes

Two main approaches are used for removal of ECs using filtration: (i) the enzyme is in the feeding solution together with the contaminants, and (ii) the enzyme is immobilized in supports (beads, membranes, gels). The immobilization of enzymes makes it possible to increase their thermostability and resistance to harsh conditions (pH, chemical reagents) and promotes easy separation from the reaction medium, allowing their reuse [77]. However, some loss of activity of the enzyme may occur due to conformational changes of the protein that makes its catalytic active sites inaccessible to the target molecule, and due to the heterogeneity of the immobilized enzyme on the support [51].

In general, enzyme immobilization processes can be grouped into five main mechanisms [78]: (i) entrapment by physical interactions into a porous matrix; (ii) encapsulation within a semipermeable barrier; (iii) adsorption through ionic forces on high surface supports; (iv) self-immobilization by cross-linking of the enzyme particles creating cross-linked enzyme aggregates (CLEAs) and cross-linked enzyme crystals (CLECs); and (v) covalent binding to a support. In (iv) and (v), the mechanism usually involves the nucleophilic attack of the coupling organic agent on the lysine moieties of the protein. The preparation of CLEAs and CLECs has several drawbacks such as low activity retention and low mechanical stability [79]. The process in (v) is preferred as it avoids leaching of the protein, which is the most expensive component in the formulated product, and it maintains a good activity.

The use of oxidoreductase enzymes in filtration can be grouped into bed filtration and membrane processes. In the current work, the degradation studies discussed only focused on the removal of ECs using enzymatic bioreactors working in a continuous flow mode, which is of practical industrial interest (

Table 8. Studies of removal of ECs by bioreactors under continuous flow mode of operation.

Enzymatic System (Source)	Type of Filtration (Mode)	EC(s) a	Enzyme Immobilization Method (Support)	Reference
Laccase (Trametes versicolor)	Fluid bed (continuous recirculation)	OP, NP	Covalent (polyacrylonitrile beads)	[80]
Laccase (Coriolopsis polizona)	Packed bed bioreactor (dead-end)	TCS, BPA, NP	Covalent (diatomaceous earth)	[81]
Laccase (Trametes versicolor)	EMR (cross-flow)	TC	Covalent (ceramic membrane)	[82]
Laccase (Trametes versicolor)	EMR (cross-flow)	Antibiotics: 12 SFAs, 6 PN, 10 FQ, 4 QN, 4 TCs	Covalent (ceramic membrane)	[83]
Laccase (Coriolopsis polizona)	Fluid bed (perfusion basket reactor, dead-end)	BPA, TCS, NP	CLEAs	[84]
Laccase (genetically modified Aspergillus oryzae)	Packed bed bioreactor (dead-end)	SMX, CBZ, DC, BPA	Adsorption (GAC)	[85]
Whole fungal cells and extracellular enzymes (P. chrysosporium)	Packed bed bioreactor (dead-end)	NPX, CBZ	Adsorption (wood chips)	[60]
Laccase (genetically modified Aspergillus oryzae)	EMR (cross-flow)	CBZ, DC, SMX, AT	-	[62]
Laccase (genetically modified Aspergillus oryzae)	EMR (cross flow)	17 non-phenolic and 12 phenolic TrOCs	-	[64]
Laccase (genetically modified Aspergillus oryzae)	EMR (cross flow)	CBZ, SMX, DC, AT, OXY	-	[86]
Laccase (T. versicolor), HRP	EMR (cross flow)	BPA	-	[87]
Crude enzyme laccase (Pleurotus ostreatus)	EMR (cross flow)	30 ECs of which 11 were PhAcs	-	[88]
Laccase (P. oryzae)	EMR (cross flow)	17 phenols	Physical entrapment	[89]
HRP, laccase (Rhus vernificera)	EMR (cross flow)	5 hydroxylated compounds	Physical entrapment	[90]
Laccase (genetically modified A. oryzae)	EMR, (cross flow)	BPA, DC	-	[91]
Mycelial pellets (Trametes versicolor)	Fluid bed bioreactor (dead-end)	CBZ	-	[61]

Notes: ^a. Abbreviations: OP: octylphenol, NP: nonylphenol, TCS: triclosan, BPA: bisphenol A, TC: tetracycline, SFA: sulfonamide, PN: penicillin, FQ: fluoroquinolone, QN: quinolone, SMX: sulfamethoxazole, CBZ: carbamazepine, DC: diclofenac, NPX: naproxen, AT: atrazine, OXY: oxybenzone, TrOC: trace organic contaminant, PhAcs: pharmaceuticals.

). The biocatalysts can be either suspended into the reaction mixture or immobilized.

5.1.1. Membrane Technologies

In the membrane processes used in enzymatic membrane reactors (EMRs), studies have focused on the use of enzymes in reverse osmosis (RO) and nanofiltration (NF) technologies. The use of microfiltration and ultrafiltration (UF) processes for EC removal is limited because the molecular weight cut-off (MWCO) of the membranes used is about several thousand Da, whereas the MW of ECs is usually less than 800 Da, which prevents the rejection of ECs by size exclusion [92]. Reverse osmosis and nanofiltration are used in the cross-flow filtration mode where the membrane configurations are in the shape of tubular and hollow fibers. However, ultrafiltration processes were also evaluated in some wastewater treatment studies with the objective of decreasing the EC concentration to environmentally safe levels, despite the presence of degraded ECs in the permeate. If the use of a membrane is in a dead-end configuration, flat sheet membranes are preferentially used.

According to the design of enzymatic reactors using membrane technologies, several configurations are used: (i) a continuously stirred tank reactor (CSTR) combined with a membrane module for product removal and a loop for recirculation of the enzyme; (ii) the membrane is submerged into the reaction tank and, in this case, the separation occurs inside the bioreactor using either hollow or flat-sheet membranes, and additional control is required for preventing cake formation and fouling; and (iii) using a module membrane with a segregation of the enzyme system within a hollow fiber membrane module, operated with feeding into the membrane lumen, and the biocatalyst is placed in the extra-capillary space [93].

When operating in continuous flow mode with membrane systems in which the enzyme is not immobilized, a periodic injection of the enzyme is required in order to maintain a constant degradation activity. This mode can operate with a recirculation loop that allows the study of the reusability of the system until a steady state with critical levels of contaminant removal is reached (Figure 2).

By evaluating the different configurations, a comparison of the advantages of using UF and NF membranes was made. Asif et al. [86] studied the degradation of four ECs with laccase using a filtration module operated in continuous flow mode without recirculation. At identical operating conditions, the removal of these ECs was in the range of 10–80% using a NF membrane formed by a polyamide thin film composite, compared with those using a UF membrane made of polyvinylidene fluoride. Similar results were obtained by extending the system to more complex matrices, including 29 ECs [64]. This higher effectiveness of NF systems is mainly due to size exclusion effects. The NF and UF membranes used in the previous studies had MWCOs of 200 Da and 30,000 Da, respectively. The MW range of all but five of the micropollutants studied was >200 Da.

On the other hand, other mechanisms may be influencing the removal of these ECs. Electrostatic repulsions with certain contaminants can be produced by charging effects if the membranes develop a negative charge at the operating pH [94]. Adsorption effects may also occur with hydrophobic contaminants, whose removal is high in the early stages of filtration but is reduced over time in the permeate due to diffusion effects in the membrane [92].

The removal of micropollutants also depends on the type of NF membrane used. The BPA removal performance of an enzymatic bioreactor coupled with a filtration module working in recirculation mode was higher when using a membrane with an active layer formed by semi-aromatic piperazine-based polyamide, compared with one with a single active layer of polyamide, due to its larger pore size, which reduced internal membrane fouling [87].

Enzymes can also induce the polymerization of organic micropollutants by oxidative cross-coupling reactions [95,96]. In this case, a gel layer generated by the accumulation of polymeric by-product macromolecules can be formed on the active surface of the membrane. This gel layer may enhance the degradation of ECs by performing as an active sorbent [88].

The use of mediators can increase the removal of micropollutants. In a bioreactor coupled with UF filtration under a continuous flow mode, the removal of BPA and DC increased in the presence of 5 µM SA (from 85 to 98% for BPA and from 60 to 80% for DC) [91]. If, in addition to the mediator, an adsorbent such as granular-activated carbon (GAC) is added, then higher degradation rates can be obtained. The addition of GAC together with SA improved the percent degradation of ECs; this was attributed to their sorption on the GAC, which promoted the interaction of the micropollutants with the active sites of the enzyme [97].

As stated earlier, the use of immobilized systems allows increased reuse of the enzyme. Physical entrapment of the enzyme in a porous membrane can be performed by the recirculation of an enzyme solution through the membrane. Through this technique, HRP-immobilized and laccase-immobilized membranes were obtained, which were able to efficiently degrade hydroxylated compounds [90]. A higher reactivity was obtained in the HRP-immobilized membrane, which was able to reduce the amount of 4-hydroxylbiphenyl by about 40% in 4 h, whereas this percentage was not reached with laccase until 24 h. In addition to the type of immobilized enzyme used, the removal of ECs also depends on the presence of EDGs that enhance enzyme activity. Lante et al. [89] studied the biodegradation of 17 phenols by laccase immobilized on a polyethersulfonate membrane by physical entrapment. These authors highlighted the importance of the type of substituent affecting the electron density of the phenol (chlorine, nitro, methoxy, alkyl groups) and its relative position.

Enzyme immobilization on supports by covalent binding requires the use of cross-linkers. The most widely used cross-linkers are glutaraldehyde and 1-ethyl-3-(3-dimethyl-aminopropyl) carbodiimide (EDC). Glutaraldehyde has two aldehyde groups that react with primary amine groups located onto the support and the enzyme forming a Schiff base, so that the cross-linker is fixed to the support by acting as a bridge [98]. EDC acts through the formation of amide bonds between primary amino groups located on the support and the carboxylic moieties of the enzyme [99]. Enzyme-grafted ceramic membranes were developed by forming a gelatin layer on the support, which was activated by treatment with a glutaraldehyde solution followed by treatment with a laccase solution [82,100]. Although these grafted membranes were not very effective in the removal of a mixture of 38 antibiotics, the addition of 10 µM of the mediator SA significantly enhanced their removal with very few exceptions (metronidazole, flumequine, and sulfanitran) [83].

5.1.2. Bed Reactors

Enzymatic complexes are also used in bed filtration (Table 8). In bed reactors, the biocatalyst can be used either fluidized or fixed inside the filters.

Glutaraldehyde was used to covalently immobilize laccase on a diatomaceous earth support that was previously activated with γ-aminoprolyltriethoxysilane (APTES) [81]. The supported laccase was tested in a packed-bed bioreactor for the removal of three endocrine disrupting chemicals (BPA, TCS and NP). The complete removal of a continuous feeding solution of 5 mg/L of each, was achieved in less than 200 min contact time using 5 g of laccase-immobilized support for BPA and TCS and 2.5 g for NP. The solid bioactivity was 0.75 U/g. In another approach, laccase was immobilized by sorption on GAC that was previously acid washed [85]. A high activity was obtained, about 750 U/g_GAC, which was equivalent to 10 mg laccase/g_GAC. In a packed-bed reactor containing 7.5 g of the enzyme–GAC complex that was operated continuously by passing single solutions of four ECs (SMX, DC, BPA, and CBZ), the EC could not be detected using 4000–8000 bed volumes (BV) depending on the EC studied.

In the specific case of BPA, the amount removed by filtration was 90% after 11,500 BV using the GAC–laccase complex versus 50% when using GAC only. The activity based on the surface of the biocatalyst was 0.58 U/m², whereas this activity in [81] for the enzyme–diatomaceous earth complex was 0.60 U/m². The high activity observed with the GAC–laccase complex may have been due to the high sorption of the pollutant on the GAC, despite a reduction in its active surface by one-third after anchoring the enzyme. The sorption of the pollutant facilitated the electron transfer from the laccase, thus promoting its degradation. This observation is supported by the much higher sorption capacity of GAC over diatomaceous earth for BPA [101].

The use of whole fungal cells was assayed in a fixed-bed reactor using a mixture of mycelia pellets of the white-rot fungus Phanerochaete chrysosporium with wood chips [60]. This allowed the degradation of ECs by a combination of extra- and intracellular enzyme systems.

The enzymatic particles can be fluidized, which promotes their contact with the solution containing the ECs. Cabana et al. [84] designed a perfusion-type basket reactor consisting of a module simulating a metal filtration membrane containing CLEAs. Through an impeller external to the module, these particles were fluidized inside the basket and interacted continuously with the solution containing the ECs, which, after being treated, was recovered by a gravity system. In the study by Jelic et al. [61], fluidization in the bioreactor was achieved through an electrovalve that generated air pulses that helped to maintain the particles in a fluidized state.

6. Future Perspectives

The experimental results and model calculations demonstrate that the purification of drinking water and wastewater utilizing degradation during filtration is likely to yield a low-cost treatment for water. It is hoped that this realization will promote experimental and modeling studies on the removal of cyanotoxins in laboratory experiments and in treatment plants dealing with lake water, which is prone to periods of harmful algal blooms. The treatment of drinking water to remove dissolved organic carbon (DOC) can also benefit from utilizing such an approach. In fact, Figure 3 in [44] indicated a 10% removal of DOC after 200 d of filtration, which cannot be explained unless degradation during the filtration occurred.

There are several perspectives to consider in order to develop a water treatment industry that operates enzymatic systems at an economically feasible level. In the first place, the commercial availability of enzymes, which could extend the variety of enzymes that can be used, could be accomplished using recombinant DNA technologies as an alternative approach to producing biocatalysts economically and on a large scale. In addition, it is necessary to develop technological innovations to reduce the costs associated with separation and purification in the production process [102]. One solution could be the use of crude enzymes or fungal cells instead of purified enzymes, which would also allow the degradation of organic compounds by means of intracellular enzyme systems, such as the cytochrome P450. However, the nutrient supply required for the use of fungal cells may promote the development of bacteria and thus introduce another source of contamination. As elaborated in Section 4.2 C, the addition of a micelle–,liposome–, or polymer–clay can secure the removal of bacteria.

Another approach to be developed is the use of immobilized multienzyme systems [103,104,105,106], which could extend the spectrum of degraded compounds. These systems could be applied both in fixed-bed treatment processes using supports that combine several enzymes, as well as in membrane systems, where they could have an additional role: the degradation of fouling precursors (e.g., dissolved organic matter) and thus extend the lifespan of the membrane.

Finally, the degradation of ECs will depend on the enzymatic activity, which, independent of the enzyme used and its source, can be enhanced by the presence of mediators at concentrations that do not increase the toxicity of the treated solution [48,83]. The mediators can be incorporated into the supporting medium containing the enzyme [107], in which case the distance and spatial distribution between the mediators and enzyme is critical.